I wanted a move that would have me working with computers. Bernard King, the Daily Mirror Group data processing manager, said that if I was seeking a job in the computer industry, I should apply for a job with Honeywell, which was challenging IBM for the commercial market in the UK. He believed that the Honeywell 200 was a real competitor to and a possible replacement for the IBM 1401 that we had bought on behalf of the Daily Mirror Group. But I did not enjoy the Honeywell interview. It was without direction. The emphasis of the available literature and wall posters was hardware control systems rather than computers, which was my burgeoning interest.
My interview with the UNIVAC division of Remington Rand, on the other hand, was quite different.1 John Woods, the UK sales and marketing manager, came across as an enthusiastic kid playing with new toys, and he showed great interest in the extent to which I had been able to pick up the new technology at the National Trade Press (NTP), including my ability to program plugboard systems and the fact I knew some (IBM 1400) NEAT programming language. Remington Rand supplied general office equipment and systems, including Remington typewriters, and Kardex filing systems and filing cabinets. Remington Shavers was somehow included in the portfolio. It made some sense that computers fitted into the division, although we had nothing to do with the established product lines. I joined UNIVAC as John’s personal assistant in the High Holborn office opposite the Old Bailey. This was a good central place to be located in 1960s London, with The Strand, Fleet Street and Covent Garden on our doorstep.
We were keen amateurs in a developing profession, but we were eager to learn. On my first morning I was sent to the West London Air Terminal, as a part of the BEACON project team installing UNIVAC 490s for British European Airways. I was to learn that UNIVAC was good at real-time computing. At the time, we convinced ourselves that we were the only computer company that could readily link terminal equipment to our mainframe computers — the terminals were teletypewriters working at 50 bits per second, very slow. I was sent to West London Air Terminal to wire as much of a UNIVAC 1004 plugboard as I could. Working on something that looked like the old-fashioned telephone exchange panels with removable connectors, we wired the computer instructions by physical wire connections rather than using software stored in the computer’s memory. I was good at this from my experience with the Daily Mirror Group. A UNIVAC 1004 was the input/output device, card reader, card punch, and 600 line per minute printer for the airline reservation system. I was very pleased that I was able to contribute.
I had seen computer rooms before, but I was not prepared for the reservation hall at British European Airways, where 256 purpose-designed and -built desks awaited the reservation clerks who would use the system. I was later to learn of the work that John Woods and Bill Dunlop had contributed to and supervised during the pre-sales work to win the British European Airways order. This had included the engineering design and construction of these purpose-built workstations. I later maintained the library of material that comprised the history of the sale as a part of my duties. The desks were quite close to one another and I was told of the design and tests that went into determining the correct strength of the spring on the seat which ensured it settled back under the terminal desk when the agent left his/her position for whatever reason. The terminal on the desk was a modified teletype with coded acetate overlays printed with the details of particular flights to give each terminal and flight its unique identity. Notches in the bottom of the overlay tripped switches to achieve flight identity, such as the London–Paris schedules. The operator read the available schedules to the telephone caller from the acetate sheet. If the flight was appropriate, the operator was able to see if there were seats available and reserve them seats while confirming passenger details. The desks were grouped by destination and area under a team leader. Today we would recognise the hall, with its public viewing gallery along one wall, as a call centre. I had never seen anything so large and dedicated to a single function.
My boss, John Woods, liked to keep on the move. Indeed, his three sales managers complained about the fact that he was always on the move — a lot of the time he was on foot somewhere in London, as he did not like using taxis. John looked like a bespectacled academic and often behaved like an impulsive schoolboy. He would roll his yellow woollen gloves into a ball and drop kick them down the office corridor to aid his progress. I did not see much of John during the first month I worked with him. I had to guess what it was he wanted me to do for him as his personal assistant. I took a lead from Bob O’Brien, who was personal assistant to the managing director, Charles W. Elliott. But it was not working.
I requested a showdown. We sat down to review a set of notes that I had prepared and John had notated: ‘y’ represented yes, and ‘n’ represented no. That was easy enough. It was ‘ω’ that was causing problems. I was treating ‘ω’ as a symbol for Woods signifying agreement for the proposed action. Our problem in communicating soon became apparent as John went through his instructions verbally: ‘yes, yes, no, bollocks, yes, no …’.
Having sorted that little difficulty out, we managed fine. I shared John’s office.
One of my jobs was the preparation of sales prices we could offer to potential customers — not the published price list, but the lowest price mandated by company policy. I worked with individual salesmen to determine the appropriate system components that would comprise the hardware we were expecting to sell. I then worked out the inter-company transfer price for hardware and software, added the cost of extras such as freight, insurance, technical support, and a cost of sales figure, to which I added the mandated 28 per cent gross profit. That was the minimum figure acceptable to the company. This price passed through management to gain approval before submission to the customer as part of a proposal. Quite often we built in an additional discount, one that might be offered during a positive negotiation. These were the days, of course, when we sold hardware, and software and support were bundled into the hardware cost. The reason we could do this was because we had no software. The salesman had a list of the engineering specifications for the components of a computer system and the availability of specific programming languages, but in the mid-1960s there was nothing as specific as application software. From the list of hardware components, the salesman had to develop a solution to a specific task or problem the potential client wanted to address. But first the salesman had to isolate what that task or problem might be. This process was called systems analysis: the job of the initial computer salesman.
Our hardware transfer costs varied by the type of system component — a new central processing unit (CPU) might have a transfer cost greater than 100 per cent, and an older model magnetic tape unit a transfer cost of 20 per cent of the sale price. I needed a 28 per cent gross margin to get to an approved selling price. It was often possible to optimise the system components, perhaps using slightly different components with advantageous transfer prices. The selling price was authorised by Roger Hunziger, the financial controller, and Charles Elliott, the managing director. I cannot remember many fights over prices. Charles Elliott was ex-army, a colonel from World War II, tall and dignified. It was rumoured that he turned off his hearing aid when meetings became long and boring.
Other front-line parts of the computer company included technical support and field engineering, which in my early days at UNIVAC were managed by John Kason, who was often at loggerheads with John Woods. I relished the day that one of the sales managers came into the office I shared with John to report that a computer language specialist was leaving the company and that his car was sitting in the garage under the building. If we did not grab the car it would be retained by the technical group. John looked at me and asked if I drove. I did, and took the car home that evening. I continued to use public transport travelling from Orpington, because the train was so convenient. It was about 18 months later when the two-year lease of the car ran out, prompting concern in the office about where it was. It was in my garage. I had made good use of the car, thank you. Possession being a determining precedent meant that my company car was replaced — and very convenient too, as it was about this time that I transferred into one of the three direct selling teams and needed a car.
The selling teams were the commercial sales team, selling to commercial customers; the government sales team selling to government, including defence; and the real-time sales team, selling systems into the commercial arena that required a terminal input and response rather than using batch input. Only UNIVAC was confident in this computing mode — or so we thought. Real-time was by definition the response needed by the user to make optimum use of the equipment we were selling. Seeing something happen almost immediately was what real-time implied.
The computer product mix was interesting. We had computer processing units with word sizes of 6, 18, 30, and 36 bits. The greater the number of bits in the word, the greater the power of the built-in machine instruction. The UNIVAC 418, as used by Intinco, for example, had an 18-bit word, and was ideal for the level of communications it was expected to achieve. The airline system at British European Airways was based upon the UNIVAC 490 family of 30-bit word machines, which required more power. The UNIVAC 1100 used 36 bits and was initially designed for scientific compute-intensive applications but was with time also used with advantage for commercial applications.
In 1965, UNIVAC and IBM were vying for dominance in the computer market. IBM had announced the 360 computer series in mid-1964 and was almost ready to start delivering on its promises. Theoretically, IBM 360 architecture meant that any program written and operated on any machine within the IBM 360 series of processing units would run on (almost) any other. IBM almost went broke with the launch of this most ambitious project, but it won the battle through superb marketing. The UNIVAC 9000 series response looked like a lame copy, using the same 8-bit byte architecture. It took a long time to mature into a saleable product.
As John Wood’s personal assistant, I had a charmed introduction into the selling environment. I was appointed account manager to International Computers and Tabulators (ICT), which became International Computers Limited (ICL) when it merged with the computer division of English Electric in 1968. When I came across them, ICT was the principal UK computer manufacturer, formed in 1959 by a merger of the British Tabulating Machine Company and Powers-Samas. In 1963, it added the business computer divisions of Ferranti, whose earliest designers included Alan Turing. It had success reselling the UNIVAC 1004 as the ICT 1004 computer system. We sold some 250 1004s to ICL in the UK, some of which ended up across the globe as ICT pursued its international ambitions. I became good friends with Bill Downing, the ICT 1004 purchasing and project manager in its Putney headquarters — it was good to get out of town occasionally. UNIVAC was the original equipment manufacturer supplying ICT, who re-badged the 1004 to complement its own product portfolio. It was good business as we also sold spare parts to ICL, but did not have to provide other technical support.
I was also assigned account responsibility for International Investment Company, Intinco, later acquired by the Daily Mirror Group and renamed International Data Highways two decades before the term ‘information superhighway’ was coined to describe the future of information communication technology. Intinco offered a service to stockbrokers through which — via teletypes in their office, connected by slow telephone lines — they could get information on the latest stock-trading prices, using the stock exchange code listing. Intinco would update the last selling price at the end of every trading day so that the price was available at the open of the next day. The service was extended to include individual and company portfolio listings and online valuations for the client. With the Daily Mirror Group acquisition of Intinco, I had the challenge of selling to my first boss, S. Chas Burt, who had appointed me data processing manager of National Trade Press, a Daily Mirror Group subsidiary, a month after I left university in 1961.
I was still very green when I went to Intinco on Clerkenwell Road for the first time and asked to speak to Bill Dunlop, the technical director. Bill took me under his wing. Indeed, I was somewhat taken aback during our initial discussion when he recommended I address, via telex, specific questions to named managers at UNIVAC headquarters in Blue Bell, Pennsylvania. I asked how he knew so much about the company. He replied that he had been the acting general manager of UNIVAC just a few months earlier. I certainly learned from Bill Dunlop that the salesman does not have to be a know-all. Bill taught me to listen, and (initially) do what he advised. As I grew in the job, and Bill’s knowledge rapidly became outdated, I was able to contribute more to the relationship, and I well remember the delight of us both when I was first able to make a sensible recommendation in my own right.
Bill Cox was the senior UNIVAC project support engineer on site. He was very tall and very intense, with a hatchet beard, and the elbows of his voluminous jumpers worn through. He programmed the operating system and rectified possible system problems via the engineer’s panel, where he could change individual bits in the instruction buffers as he went along. This was great in the short term, but it was all done too fast to be documented and caused repeated headaches when we updated the software operating system and re-introduced the previously corrected errors.
We had a team of engineers on site for two shifts to cover evening data entry and the continual system tinkering that went on. Etele Hustie, a small, modest engineer who was a refugee from the Hungarian uprising of 1953, was the chief engineer. Etele and I became friends and shared many weekends at Intinco installing additional equipment, closing the main road on many Sundays as we had to lift the heavy equipment into the building by overhead crane. At Intinco we always had the very latest in mass storage devices — in today’s terms the storage was minuscule, measured in thousands of words; we were not into mega-anything at this stage. The UNIVAC technologies were drum oriented and it was to be 10 years before we started to see disks as we know them today.
We had a few panics sustaining the online system, one of which certainly sticks in the mind. I got into the Holborn Viaduct office at the normal time of 9.00 am to be told that Intinco was down, had been down all night, and that Bill Dunlop was apoplectic. I got to Clerkenwell Road within minutes to find all of Intinco management in the operations manager’s office looking through the observation window at Etele. He was skipping rope.
I went into the computer room through the engineer’s office where Etele’s staff were cowering, poised to take spare parts to Etele when called upon. I ventured into the computer room and spoke softly to Etele as I approached him from behind.
‘Hi Etele, it’s Chris. What can I do to help?’
He stopped his exercise and, turning, said:
‘Thank goodness you are here. Keep that lot off my back. I cannot think when they keep interrupting, every two minutes, to ask when the fault will be fixed.’
‘OK. I can do that. But what’s with the skipping?’
‘A moving wire is the only way I can keep Bill Dunlop and Cliff Griffith away from me, and I have told them it helps me concentrate.’
Needless to say, Etele fixed the fault within a reasonable time and no major harm was done. Etele was my kind of engineer. When the NASA space program used a dual UNIVAC 418 system as a part of its world-wide network from an office in Holborn, Etele and his lads were on standby during critical space missions.
I stayed with the BEACON project as John’s representative until completion. The BEACON project team was an interesting mix of experts. The senior members were Phil Fellows, an American with airline experience; deputy project manager John (‘um-yeah’) Harrison, a fast-talking chain-smoker who was recorded at one presentation uttering ‘um-yeah’ 30 times in one minute; Ed Mack, an American albino who was the programming genius who made it all happen and worked closely with Brian O’Heron (the two later joined forces in the US to develop computer medical systems); and Dave Phillips, a voluble Welshman who worked on planning and scheduling aspects of the project.
Phil Fellows’s number one rule on the project was the integrity of program interfaces. He kept repeating that the only reason for instant dismissal from the project would be the failure to gain agreement for, and documentation of, an interface change. Once he even added: ‘Not even sex on the computer console will constitute a reason for dismissal, provided the project is not inconvenienced.’
Of course, the challenge was accepted. We waited anxiously for the outcome of that very act enjoyed by Ian Douglas and Pam Sainsbury, who were discovered by British European Airways security staff overnight, when they were ‘working over’. No disciplinary action was taken — only notoriety for the perpetrators.
Another overnight activity remains a mystery to this day: the 1004 printer would burst into life and print reams of gibberish for no known reason. This happened a few times. The attendant computer operator was not able to cancel the print file and fed boxes of continuous stationery until the system had disgorged itself. These files became affectionately known as the ‘Dead Sea Scrolls’ and were recorded as such.
The BEACON project team became the founding members of the UNIVAC London Real-Time Research and Development Centre later established in Paddington. There they continued to develop airlines systems as a part of the UNIVAC Standard Airline System, which helped the company to share that particular market segment with IBM for many years.
David Streeton, an engineer who understood the technology we were selling, managed the real-time sales group. I always felt that his team had the most fun and were the most technically-competent selling group the company had. I was later fortunate enough to join this group.
As a group we had considerable success, and tended to play too hard. A crestfallen John Oswald came into work one morning dressed in the same clothes he had worn to work — and the pub — the day before. He was late home. His wife took exception to his lateness, and believed she could smell perfume on John’s suit, so she threw him out of home. As he tried to sleep in his car, in the driveway, he was kept awake as his possessions were thrown on the front lawn. Thankfully, there was no lasting drama — he got into his home the next evening and we understand nothing else was said.
Brian Lawrence, an old-school engineer, and I took Ed Mack to meet the Trustee Savings Bank in Altrincham, Cheshire, for its presentation to explain a request for proposal it was going to issue. We decided to drive up. Ed started telling rude jokes from the moment he got into the car, continuing until we found the hotel in Altrincham, and then on the way back. The bank’s plans were innovative and Ed gave another virtuoso performance in assuring them that Sperry had the hardware, software, and the plans to meet its every wish. Bob Brotherton, head of data processing at Trustee Savings Bank, explained to us the bank’s relationship with the Bank of England, which would be influential in any final decision. We submitted a proposal, which Mr Brotherton recommended, but was overruled, and the bank acquired an ICL System 4 configuration. It would take Mr Brotherton 14 years to win approval for a Sperry 1100 system.
One of the reference accounts that David had sold was to Rediffusion Television, where a six-bit word commercial system was used for real-time advertisement booking and scheduling. The UNIVAC 1050 was not really built for real-time operation, but David’s design and Bill Chatham’s technical management worked wonders.
David Streeton did not need to use his team very much when he sold the Corporation for Economic and Industrial Research (CEIR) a large mainframe for scientific processing. Initially, Mike Seaton was the assigned CEIR salesman. But Mike was too precious and thin-skinned, flouncing into David’s office and refusing to deal with CEIR because they were not nice to him. So David was obliged to run with the account.
Eventually UNIVAC moved the real-time group within Remington House, and David was relegated to a windowless office that he did not like and which left his team in an open-plan environment. Not to be defeated, our David spent a weekend wallpapering his cubbyhole with wood-grain pattern wallpaper and installing a fridge, which ensured he still got visitors. For the open-plan office, he installed an early model coffee percolator and we learnt, over time, not to need six coffees each morning and afternoon.
We all worked on the sale of the Sperry 1100 system to Shell International Petroleum Company. Shell had been using the Computer Services Bureau in Birmingham for linear programming (the logical solution of multiple simultaneous equations), and our solution introduced remote job entry to its UK operation. Our remote job entry equipment was a UNIVAC 1004, the same hardware we had been selling directly through ICL, as a small commercial system, at sites geographically separate to the main computer centre and connected by telephone line to the main computer. We were still working with punch cards, although the remote job entry system could include magnetic tape storage. Ian Meeker, who was rather pompous, carrying a tightly rolled umbrella that matched his bowler hat, nominally headed the sales effort. He used the Shell Pecten symbol as inspiration for the project name: ‘poly-entrant computing through (the) entire network’.
As always seemed to be the case, Ian and I worked at the photocopier all night before the proposal was due. We were still sorting pages and hole punching them for insertion into three-ring binders on the back seat of the taxi as we drove to Shell Centre to meet the tender submission deadline.
We won the bid. Ian moved out of our open-plan office at Remington House and into an office at Shell Centre, despite the fact that the lead-time for delivery of the 1100 was 24 months away. Ian also employed his own secretary at Shell. He established a solid relationship with Norman Rosenthal, the Shell manager for the project. Everything worked, and eventually David Streeton demanded that Ian return to Remington House.
Ian arrived on a Monday morning at home office and we could hear an almighty row coming from David’s wallpapered enclave. Ian Meeker’s belief and David’s denial that the ‘sale to Shell was enough to guarantee his position for life’ caused Ian to storm out. An equally angry but determined David Streeton stalked into the open-plan office and announced to his expectant audience:
‘You are all promoted to senior sales executives. Ian Meeker’s excuse for not wanting to come back from Shell Centre is that he refuses to have to share any office with “junior” sales executives! I have told him he will not have to.’
The team accepted that. Funny how that decision has permeated the whole industry. You would only ever meet senior sales executives in the computer sector from that day on.
It was inevitable that Ian would move on. He joined the international division of the company and worked in Eastern Europe — we presumed that he would still have his bowler hat on, even in Russia. He was attending to Aeroflot business when he died suddenly in Moscow.
With Shell and British Petroleum (BP) as customers, it was natural that the team would focus its attention on the tender request put out by Shell-Mex and BP Ltd, the marketing company for Shell, BP, National Benzole and Power Petrol. Norman Blatch was the lead salesman. Large and almost larger than life, Norman would brighten even a dull day with his enthusiasm and barrow boy, finger-flicking mannerisms when making a point. Norman was ecstatic when we won the job, but we lost regular contact with him as he devoted himself to setting up the project. The Shell-Mex and BP Ltd representatives were a charming crowd. They killed a bottle and a half of scotch with Norman and David in David’s office one afternoon. The session ended with the unanimous acceptance of the fact that Shell-Mex and BP Ltd people compared very well with the more ruthless staff we dealt with at Shell and BP.
I think the team was jealous of Norman. We were all formally invited to dine with Norman at a restaurant in the Aldwych one Saturday evening to celebrate the installation of the first Shell-Mex and BP computer system at their Hemel Hempsted computer centre, north of London. Norman had the reputation for being a bit stingy, so the lads met for a pint or two before going to the restaurant where we dined in a private room. Norman had also invited along Shell-Mex representatives and the team mood became blacker as we were suddenly obliged to have to work for our supper and be nice to our new customers. But at the conclusion of the meal, each member of the team was presented with gold engraved cufflinks as a thank you for helping during the sales process. That humbled us. I still wear those cufflinks.
Our specialisation in real-time computing led to a request from UNIVAC Scandinavia, hoping to sell a UNIVAC 418 system to a prospect in Lapland. Tjaereborg Rejser A/S was a travel agent and airline owner, and a user of an early Control Data Corporation (CDC) 3300 system. Per Krogiger, the Tjaereborg pastor, had initially started a small business taking his parishioners on bus and coach trips. The business had grown very large, and had offices in the Scandinavian capitals and a booking and reservation system for Stirling Airlines, also owned by the pastor. He also operated a conventional travel agency business. It was a fascinating trip, and I got to travel in a sled pulled by reindeer within the Arctic Circle. But more significant was a lesson in pragmatism and a new definition of ‘real time’ from the pastor. He told us that the response times from the CDC system were:
Tjaereborg Town Office: 30 seconds
Copenhagen: 10 minutes
Helsinki: 20 minutes
Oslo: 30 to 40 minutes
Foolishly, we decried these as meaningless response times and argued that people would not wait this long. Per Krogiger knew better than we did. The Copenhagen office was in a coffee shop, where the waiting public sampled the coffee; the Helsinki office was associated with a newsagents and book shop; and the Oslo office was based in a department store. Per Krogiger certainly provided a complete service and made money at every turn. He did not need response times under three seconds to be satisfied with his installed system.
The government sales group found it difficult to penetrate and win government business, which was given to ICL, the only UK computer maker, as government policy. One day we received a request for tender from the Royal Navy for a system to handle the logistics of the Polaris submarine system. The request called for two computer centres, one in Bath and one in Darlington. Was it worth UNIVAC bothering to tender? We discussed this at length. We decided to bid. We knew we would not win the business, and so gambled upon making the expected sale as unprofitable for ICL as possible. We put in a bid, using the latest technology to further test ICL, at 50 per cent of the published price list. We were also very generous in our support offer. The bid was forgotten.
After several months interval and no prior indication, we received notification from the navy that we had won the business. A quick run through of the figures showed that there was no way we would be able to show the necessary 28 per cent gross margin. Worse, we had lodged no paperwork with the company to state that we had bid, and certainly had no approvals, at the winning price. We brainstormed long and hard about what to do.
Eventually we decided that we would have to cheat the system. We fabricated the paperwork that should have been produced months earlier. Once this course of action was determined it should have been simple, but it was not. The company name had changed and for a while we could not find any of the old stationery. The pre-dated document looked awfully pristine. It was aged under the mat just inside John’s office for half a day and surreptitiously lodged in Charles Elliott’s office, at the bottom of secretary Rona Cook’s in-tray.
That afternoon we marched into Charles’s office with the good news and the order letter from the navy. It was never publicly disclosed how hard he and Roger Hunziger had to fight to get approval for the navy business from UNIVAC headquarters, but they managed it.
But the navy saga did not stop there. Out of the blue, Dennis Bizeray, who managed the government sales team, sent John Woods a note, the intent of which was to deprive Mike Norman of his rightful commission. Dennis argued that Mike did not deserve the full commission and that he was too young to be able to handle the responsibility of so much money. Fortunately, common sense prevailed and Mike got his money. But he did not stay much longer with the company.
Yes, there were times when we were cowboys.
We could not have enjoyed the success we had without the help and constant attention of our secretaries. Jane Newsome and Sue Dixon were great and, if anything, put in even longer hours than we did. They were in the team and were included wherever possible in social as well as business functions.
Jane and Sue came into their own at proposal time. As salespeople, we were looking for the opportunity to bid systems to prospective customers. We would do our very best to make sure we wrote the requirements contained in the request for tender on our prospect’s behalf if at all possible. This did not happen that often. The date of submission for a proposal has a life of its own. Initially, we would have plenty of time and when compiling the proposal schedule we would allow for slippage, anticipating that it would not be needed. But it always was, and last-minute panic was the norm. These were the days when reproduction of a number of copies was achieved by using a Roneo-Vickers reproducing machine. Pages were typed onto ‘skins’ that took the impression from the electric typewriter through which the ink would permeate on to plain paper. If the impression was too hard (‘o’s were notorious), the centre of the letter would fall out and the resultant print image would be a filled-in letter. The skins could be corrected using a product called Snopake. We were able to change the skins a few times in an attempt to reuse previous work, but always got into trouble with the pricing section. This would be changed so often, as the proposal developed, that the layers of the red-coloured corrections were multiplied until the offending numbers were positioned on blisters on the skin. As the proposal neared completion, every part of the organisation wanted to add its thoughts. Quite often, the print reproduction was severely prejudiced by the thickness of the skin and the actual prices were almost indecipherable on the finished product. We used three-ring binders and I can remember us submitting three copies of a set of more than 20 volumes. Manuals were photocopied more often than not, and there were many overnight efforts to meet submission dates.
The chain-smoking Ann Whiting was our advertising and public relations manager. Ann was a champion. She did the UK subsidiary great service in that we were able to invite Grace Murray Hopper to London as a guest. Grace was one of the acknowledged pioneers of computing from the Eckert and Mauchly Electrical Research Laboratories and the University of Pennsylvania from which originated, under the Remington Rand banner, the UNIVAC 1 (UNIVersal Automatic Computer 1), arguably the world’s first computer. When I first heard Lieutenant Commander US Navy (Reserve) Murray Hopper speak, she was already over 60 years old, severely dressed in her navy uniform. She carried an enormous handbag out of which she would produce the tools of her trade, as well as her knitting. To illustrate her talk she showed the speed of the computer through a 30 cm long piece of wire. This was the length of wire that, at the speed of light, represented one nanosecond: one millionth of a second. Even then, we were working in nanoseconds. Grace was eventually promoted, in her 70s, to be a Rear Admiral in the US Navy and was made a Fellow of the British Computer Society.
Our customers were always keen to receive such famous visitors and I was fortunate enough to accompany her on several visits and have time for one-on-one discussions over lunch. She loved the Connaught Rooms off the Aldwych for afternoon tea.
Figure 2.1: Grace Murray Hopper as I remember her.
Source: Wikipedia Commons.
Training did not seem to be an issue, but I do remember Mike Lewis from the Shell project team teaching us the 1100 assembler language on Monday evenings. I was also extremely lucky to be able to attend the first European project management course, which was run from the company’s European headquarters in Zurich. I was the only UK student in attendance. The first two weeks involved daily visiting lecturers from the British European Airways BEACON airline project. I suspect we were taught ‘ideal’ project management, as it included sessions on pricing and public relations, items that were not generally the UNIVAC project manager’s business. We were also told that UNIVAC expected its programmers to be able to produce six lines of computer code per day. This included the writing, testing, and documentation of the code as it stood, as well as its place in the system. Six lines of code per day did not seem very much, especially when we knew that an airline reservation with all the checking that takes place could be around 30,000 lines of code long. In simple terms, it would require two-and-a-half man years of effort to produce one reservation, and this was using the most effective software code, the code designed for the computer upon which it was to run. This fact was the rationale for UNIVAC to write a generalised software system, for airline applications, that might be sold to many different customers. Also, if the airline itself decided to write a generalised system that was capable of adaption, it would eventually become a software competitor to UNIVAC. Both UNIVAC and IBM set themselves the airline business as a target and each would achieve approximately a 50 per cent share of the market.
It did not matter how good a system UNIVAC had implemented at British European Airways: when it was absorbed into the British Overseas Airways Corporation there was going to be a competition between BEACON and BOADICEA (the British Overseas Airways Digital Information Computer for Electronic Automation system that was running on large IBM platforms). The data processing director of BOADICEA was Peter Hermon. We knew Peter Hermon at UNIVAC, as he had been a prospect when he worked at Dunlop Engineering in Birmingham. He had never been an easy person to approach, and we feared the inevitable face-off. Peter always preferred IBM, but for the British Airways system we could also expect that the UK Government would bring pressure to bear for them to consider ICL, the indigenous supplier. ICL by this time had absorbed English Electric Computers whose genesis had been LEO Computers, developed for Lyons Tea Houses.
As occurred 30 years later when Qantas took over Australian Airlines, the technical, independent view was that the Sperry+UNIVAC system was the most comprehensive and advanced of available facilities. In both these cases, the decision was made to consolidate on the IBM platform, as operated by the major airline partner.
The UNIVAC division of Sperry Rand increasingly became the dominant division of the company and we enjoyed almost annual name changes to reflect the changes in the conglomerate. Other divisions included the original Remington division, with its Kardex filing systems, and Remington typewriters and shavers. We were given Remington shavers at Christmas for distribution to our best customers as gifts, which were always welcome. The New Holland division made and distributed a range of heavy farm machinery. The New Holland name was better known in rural communities than UNIVAC or Sperry. Frank Moore sold a computer system to the Sperry Gyroscope Company in Bracknell, Berkshire, which became another reference site for us, as most technical people were aware of the Sperry Gyroscope connection.
We anticipated and learned to live with annual organisational changes, and in the sexy industry in which we were working, jobs were not hard to come by. At the start of the 1969 fiscal year, the company decided to change the structure of the sales organisation to a purely geographic structure. Sales teams were reassigned, and Midlands (Birmingham), and Northern Region (Altrincham) offices were established. There were many changes at the top, we had an influx of recruits across the company, and UNIVAC House was acquired above Euston Station to house the growing community.
One project that the new London region sales group undertook was fascinating. EMI Records worked with us to evaluate the potential of a real-time manufacturing system that would control the day’s production of records to reflect the wholesale orders it had received that day, rather than it making guesses as to what might be ordered for storage prior to distribution. With the system we proposed, we were looking to coordinate the actual manufacture of vinyl records through the merging of many conveyor belts from the multiple vinyl presses available, so that a particular wholesaler’s order would arrive at dispatch at the appropriate time. The mathematics were horrendous. As UNIVAC salespeople, we knew the rudiments of queuing theory to be able to explain to our prospective customers how the interrupt processing of our equipment worked to handle communications, but we were all stretched coordinating the travel of records down numerous conveyor belts. The system was never built, but I was able to use skills acquired from this exercise when I got to Control Data Corporation and became involved in process and production control.
They were busy, interesting times, and a lot of fun.
1 UNIVAC is an abbreviation of UNIVersal Automatic Computer, referring to a line of electronic digital stored program computers. The UNIVAC 1 (delivered in 1951) was the first commercial computer produced in the US and the BINAC built by the company was the first general-purpose computer for commercial use. After I joined the company, it went through a number of name changes, while still wanting to retain the name UNIVAC and its association with the first commercial computer ever sold. In 1965, the computer was in its second generation, having moved from valve technology to solid state electronics.